Article.py

import wikipediaapi

class Article:
    def __init__(self):
        self.article = None
        self.article_data = []
        self.id_counter = 0

    def set_summary(self):
        if self.article.summary:
            for text in self.article.summary.split('\n'):
                self.id_counter += 1
                self.article_data.append(
                    {
                        'id': self.id_counter,
                        'section': 'Summary',
                        'text': text.lower()
                    }
                )

    def set_sections_and_texts(self, sections):
        for section in sections:
            if section.text:
                for text in section.text.split('\n'):
                    self.id_counter += 1
                    self.article_data.append(
                        {
                            'id': self.id_counter,
                            'section': section.title,
                            'text': text.lower()
                        }
                    )
            if len(section.sections) > 0:
                self.set_sections_and_texts(section.sections)

    def clean_data(self):
        unwanted_sections = ['See also', 'External links']
        cleaned_data = []
        for data in self.article_data:
            if len(data['text']) > 1 and data['section'] not in unwanted_sections:
                cleaned_data.append(data)
            
        self.article_data = cleaned_data

    def get_article_data(self, article_name):
        self.article = wikipediaapi.Wikipedia('en').page(article_name)

        if not self.article.exists():
            return []
        else:
            self.set_summary()
            self.set_sections_and_texts(self.article.sections)
            self.clean_data()

        return self.article_data